Show HN: Power-Performance Curves for CPUs

[lumip]

Hi HN!

Following the release of the latest generation of CPUs by AMD and Intel with rather high power requirements and the corresponding uptake in interest of undervolting / power level adjustment, I was interested to find out in more detail how these CPU models would perform and compare at different power settings. However, it couldn't find an online source that would provide a satisfactory amount of data to be really useful to compare different models. So I took what data I could find to set up a Bayesian regression model.

Find the results and more details on the linked page. Hope this is useful or of interest. Any feedback is welcome.

Especially great would be if someone knows a source for or would be able and willing to contribute data for more CPU models (especially Intel 13th Gen.).

PS.: Please not that I haven't been able to spend much time on optimizing the website.

[btdmaster]

Reading through "The Predictive Model",

Are the parameters a and c positive? I can only get monotonic increasing with decreasing gradient with a and c both negative.

[lumip]

Good catch, those are typos. Parameters a and c are both negative, as you say. I'll have to fix that description, thanks!

[BSVogler]

Very interesting. Thanks for sharing.

You assume that the function is concave. What is the reason for that? I could imagine that the real function is monotonic but non-strict, hence non-concave.

[lumip]

Thanks! That's a good question and it is, of course, just an assumption.

My main reasoning was that as we increase the power level more and more, I would expect diminishing returns, i.e., a saturation curve as we would see in many physical (and especially electromagnetical) processes. Hence concavity. I would find a relation where we could get essentially limitless compute out of the die if we just provide enough power much less likely.

I could be wrong about that, of course. It could be that the function is not (globally) concave and just has a cut-off at some upper limit (basically when the CPU blows up). Or maybe the real function is concave but doesn't follow the exponential that I used here.

In the end I settled at the current formulation somewhat out of convenience and because it seemed to be able to explain/fit the (limited) data already quite well.

I did some experimentation using concave (again!) polynomials, but found the that in that case it was much more difficult to achieve convergence - and when it did, the results weren't much different than the current ones.

Do you have any particular suggestion for an alternative model in mind? I will probably continue experimenting with this a bit more in the future.

[ttfkam]

A shame Apple silicon isn't in that mix for reference (though I realize how hard apples-to-apples testing would be, no pun intended).

[lumip]

That would indeed be a nice addition. Unfortunately I don't have any such devices at hand. I'm not even sure if their power targets can be adjusted, but even if not it would be worthwhile to include them at their reference/design power.

I'd also have to check whether the same benchmark is available for MacOS, but fusing scores from multiple benchmarks for this would be an interesting long-term goal of mine anyways.

(Thinking about the first point, I should probably highlight the design power point for each model somehow in the plot).

[brokenmachine]

Very useful. Thanks for that.

[aver4geredditor]

Very cool, thanks for posting